kdu_block [struct]

Records the dimensions of the code-block. These are the true dimensions, used by the block encoder or decoder — they are unaffected by regions of interest or geometric transformations.

Identifies the region of interest inside the block. This region is not corrected for any prevailing geometric transformations, which may be in force. region.pos identifies the upper left hand corner of the region, relative to the upper left hand corner of the code-block. Thus, if the region of interest is the entire code-block, region.pos will equal (0,0) and region.size will equal size, regardless of the geometric transformation flags described.

Together, the transpose, vflip and hflip fields identify any geometric transformations which need to be applied to the code-block samples after they have been decoded, or before they are encoded. These quantities are to be interpreted as follows:

• During decoding, the block is first decoded over its full size. If indicated region is then extracted from the block and transformed as follows. If transpose is true, the extracted region is first transposed. After any such transposition, the vflip and hflip flags are used to flip the region in the vertical and horizontal directions, respectively. Note that all of these operations can be collapsed into a function which copies data out of the region of the block into a buffer used for the inverse DWT.
• During encoding, the region of interest must necessarily be the entire block. The vflip and hflip flags control whether or not the original sample data is flipped vertically and horizontally. Then, if transpose is true, the appropriately flipped sample block is transposed. Finally, the transposed block is encoded.

See description of transpose.

See description of transpose.

Holds the logical OR of the coding mode flags. These flags may be tested using the Cmodes_BYPASS, Cmodes_RESET, Cmodes_RESTART, Cmodes_CAUSAL, Cmodes_ERTERM and Cmodes_SEGMARK macros defined in "kdu_params.h". For more information on the coding modes, see the description of the Cmodes attribute offered by the cod_params parameter class.

Holds one of LL_BAND, HL_BAND (horizontally high-pass), LH_BAND (vertically high-pass) or HH_BAND. The subband orientation affects context formation for block encoding and decoding. The value of this member is unaffected by the geometric transformation flags, since it has nothing to do with apparent dimensions.

Encourages a block decoder to attempt to locate and conceal errors in the embedded bit-stream it is decoding. For a discussion of how this may be accomplished, the reader is referred to Section 12.4.3 in the book by Taubman and Marcellin.

For more information regarding the interpretation of the resilient mode, consult the comments appearing with the declaration of kdu_codestream::set_resilient.

Encourages a block decoder to check for compliance, generating an error message through kdu_error if any problems are detected.

For more information regarding the interpretation of the resilient mode, consult the comments appearing with the declaration of kdu_codestream::set_fussy.

Maximum number of magnitude bit-planes for the subband, including ROI adjustments. See kdu_subband::get_K_max_prime for further discussion of this quantity.

Written by the encoder; read by the decoder. For an explanation of this quantity, refer to Section 8.4.2 in the book by Taubman and Marcellin.

Written by the encoder; read by the decoder. Number of coding passes generated (encoder) or available (decoder), starting from the first bit-plane after the initial missing_msbs bit-planes.

Entries are written by the encoder; read by the decoder. The first entry holds the total number of code bytes associated with the first coding pass of the first bit-plane following the initial missing_msbs empty bit-planes. Subsequent entries identify lengths of the incremental contributions made by each consecutive coding pass to the total embedded bit-stream for the code-block.

Use max_passes and set_max_passes to make sure this array contains sufficient elements.

The entries in this array must be strictly decreasing, with the exception that 0's may be interspersed into the sequence.

• When used for compression, the entries in this array usually hold a suitably normalized and shifted version of log(lambda(z)), where lambda(z) is the distortion-length slope value for any point on the convex HULL of the operational distortion-rate curve for the code-block (see Section 8.2 of the book by Taubman and Marcellin for more details). Zero values correspond to points not on the convex HULL. The final coding pass contributed by any given code-block to any quality layer can never have a zero-valued slope.
• When used for transcoding, or with interchange codestreams (i.e., when the internal code-stream management machinery was created using the particular form of the overloaded kdu_codestream::create function which takes neither a compressed data source, nor a compressed data target), the entries of the pass_slopes array should be set equal to 0xFFFF minus the index (starting from 0) of the quality layer to which the coding pass contributes, except where a later coding pass contributes to the same quality layer, in which case the pass_slopes entry should be 0. For more details, consult the comments appearing with the definition of kdu_codestream::trans_out and kdu_precinct::get_packets.
• When used for input, the pass_slopes array is filled out following exactly the rules described above for transcoding.

Use max_passes and set_max_passes to make sure this array contains sufficient elements.

This array is allocated in such a way as to allow access to elements with indices in the range -1 through max_bytes-1. This can be useful when working with the MQ encoder.

Use max_bytes and set_max_bytes to make sure this array contains sufficient elements.

Current size of the pass_lengths and pass_slopes arrays.

Current size of the byte_buffer array.

Not used by the internal code-stream management machinery itself, this array provides a convenient shared resource for the benefit of block encoder and decoder implementations. Use max_samples and set_max_samples to make sure the array contains sufficient elements.

Not used by the internal code-stream management machinery itself, this array provides a convenient shared resource for the benefit of block encoder and decoder implementations. Use max_contexts and set_max_contexts to make sure the array contains sufficient elements.

Current size of the sample_buffer array.

Current size of the context_buffer array.

True if a warning message has already been issued by the block decoder when operating in resilient mode — saves a massive number of warnings from being delivered when a heavily corrupted code-stream is encountered.

True if a warning message has already been issued by the block encoder to indicate that the 32-bit implementation has insufficient precision to represent ROI foreground and background regions completely losslessly.